Modulations of local synchrony over time lead to resting-state functional connectivity in a parsimonious large-scale brain model

Abstract

Biophysical models of large-scale brain activity are a fundamental tool for understanding the mechanisms underlying the patterns observed with neuroimaging. These models combine a macroscopic description of the within- and between-ensemble dynamics of neurons within a single architecture. A challenge for these models is accounting for modulations of within-ensemble synchrony over time. Such modulations in local synchrony are fundamental for modeling behavioral tasks and resting-state activity. Another challenge comes from the difficulty in parametrizing large scale brain models which hinders researching principles related with between-ensembles differences. Here we derive a parsimonious large scale brain model that can describe fluctuations of local synchrony. Crucially, we do not reduce within-ensemble dynamics to macroscopic variables first, instead we consider within and between-ensemble interactions similarly while preserving their physiological differences. The dynamics of within-ensemble synchrony can be tuned with a parameter which manipulates local connectivity strength. We simulated resting-state static and time-resolved functional connectivity of alpha band envelopes in models with identical and dissimilar local connectivities. We show that functional connectivity emerges when there are high fluctuations of local and global synchrony simultaneously (i.e. metastable dynamics). We also show that for most ensembles, leaning towards local asynchrony or synchrony correlates with the functional connectivity with other ensembles, with the exception of some regions belonging to the default-mode network. Author summaryHere we present and evaluate a parsimonious model of large-scale brain activity. The model represents the brain as a network-of-networks structure. The sub-networks describe the neural activity within a brain region, and the global network encodes interactions between brain regions. Unlike other models, it capture progressive changes of local synchrony and local dynamics can be tuned with one parameter. Therefore the model could be used not only to model resting-state, but also behavioural tasks. Furthermore, we describe a simple framework that can deal with the arduous task of identifying global and local parameters.